A 3D mesh convolution-based autoencoder for geometry compression

2026-03-02

AI Summary

This work proposes an end-to-end geometry compression method for arbitrary irregular 3D meshes that requires no preprocessing and makes no assumptions about manifoldness or watertightness. Built upon a 3D mesh convolutional autoencoder, the approach employs specially designed pooling and unpooling operations to compress the input into a compact base-mesh latent space while preserving topological connectivity, enabling high-fidelity geometric reconstruction. As the first compression framework capable of directly handling arbitrary unstructured meshes, the method significantly outperforms existing techniques across multiple datasets, achieving not only superior reconstruction accuracy but also enhanced semantic expressiveness and comparability in the latent space.

Technology Category

• Machine Learning: Deep Generative Models & Autoencoders
• Computer Vision: 3D Computer Vision
• Knowledge Representation and Reasoning: Geometric, Spatial, and Temporal Reasoning

Application Category

• Graph Algorithms and Modeling for the Web: Graph embeddings and representation learning for Web-related graphs
• Semantics and Knowledge: Methods, algorithms and applications for the development of semantic models, knowledge graphs and other forms of structured data models with machine-interpretable semantics
• Economics, Online Markets and Human Computation: Architectures and workflows that use LLMs for crowd work

Abstract

In this paper, we introduce a novel 3D mesh convolution-based autoencoder for geometry compression, able to deal with irregular mesh data without requiring neither preprocessing nor manifold/watertightness conditions. The proposed approach extracts meaningful latent representations by learning features directly from the mesh faces, while preserving connectivity through dedicated pooling and unpooling operations. The encoder compresses the input mesh into a compact base mesh space, which ensures that the latent space remains comparable. The decoder reconstructs the original connectivity and restores the compressed geometry to its full resolution. Extensive experiments on multi-class datasets demonstrate that our method outperforms state-of-the-art approaches in both 3D mesh geometry reconstruction and latent space classification tasks. Code available at: github.com/germainGB/MeshConv3D

Problem

• 3D mesh
• geometry compression
• irregular mesh
• autoencoder
• mesh connectivity

Innovation

• 3D mesh convolution
• geometry compression
• autoencoder
• irregular mesh processing
• topology-preserving pooling